Abstract: A Multi-Quadric orthogonal rotational regression design is used in the quantitative synthetic multi-factor and multi index study of dryland wheat production in Loess Plateau.Based on the data of field experiments in two years,mathematical models are developed by using a mini-computer.The quantitative intensities of each factor with respect to its multi-objective functional indices are expiated by mathematical analysis and the best combination of whole set measures for high yeilding is selected by simulation.Experiment result proves that not only the mixed application of nitrogen,phosphorous and farm manure(FM)is necessary,but also the optimum ratio of these nutients is important to fulfill the desired yield in dryland wheat.Nitrogen fertilizer has greater potential for increasing yield in normal rainfall years,so while keeping the usual application of phosphates,nitrogen should be increased;while in drought years,yield will be decreased with over-application of nitrogen fertilizers.Therefore,with the normal amount of nitrogen,addtional application of phosphorous is required.It is concluded that under present conditions an feasible way for improving wheat yield in dryland is a combination of optimum population with optimum seeding date and rate,increasing the fertilizing efficiency by using inorganic fertilizers and FM,and improving water utilization.

Key words: Dryland wheat, Combination design, Mathematical model